From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <rahul.lakkireddy@chelsio.com>
Received: from stargate3.asicdesigners.com (stargate.chelsio.com
 [67.207.112.58]) by dpdk.org (Postfix) with ESMTP id 834F3C2FC
 for <dev@dpdk.org>; Thu,  9 Jul 2015 11:24:58 +0200 (CEST)
Received: from localhost (scalar.blr.asicdesigners.com [10.193.185.94])
 by stargate3.asicdesigners.com (8.13.8/8.13.8) with ESMTP id t699Oq0Z032092;
 Thu, 9 Jul 2015 02:24:52 -0700
Date: Thu, 9 Jul 2015 20:24:35 +0530
From: Rahul Lakkireddy <rahul.lakkireddy@chelsio.com>
To: Thomas Monjalon <thomas.monjalon@6wind.com>
Message-ID: <20150709145433.GA3409@scalar.blr.asicdesigners.com>
References: <cover.1436288467.git.rahul.lakkireddy@chelsio.com>
 <96172d6f608d59f6d8463d407f21d08a963d6d5a.1436288467.git.rahul.lakkireddy@chelsio.com>
 <6332495.cKaf3cHUcn@xps13>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <6332495.cKaf3cHUcn@xps13>
User-Agent: Mutt/1.5.23 (2014-03-12)
Cc: dev@dpdk.org, Felix Marti <felix@chelsio.com>,
 Nirranjan Kirubaharan <nirranjan@chelsio.com>,
 Kumar Sanghvi <kumaras@chelsio.com>
Subject: Re: [dpdk-dev] [PATCH 1/3] cxgbe: Fix RX performance for cxgbe PMD.
X-BeenThere: dev@dpdk.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: patches and discussions about DPDK <dev.dpdk.org>
List-Unsubscribe: <http://dpdk.org/ml/options/dev>,
 <mailto:dev-request@dpdk.org?subject=unsubscribe>
List-Archive: <http://dpdk.org/ml/archives/dev/>
List-Post: <mailto:dev@dpdk.org>
List-Help: <mailto:dev-request@dpdk.org?subject=help>
List-Subscribe: <http://dpdk.org/ml/listinfo/dev>,
 <mailto:dev-request@dpdk.org?subject=subscribe>
X-List-Received-Date: Thu, 09 Jul 2015 09:24:59 -0000

Hi Thomas,

On Tue, Jul 07, 2015 at 23:30:38 +0200, Thomas Monjalon wrote:
> 2015-07-07 22:42, Rahul Lakkireddy:
> > CXGBE PMD rx allocates a new mbuf everytime, which could lead to performance
> > hit.  Instead, do bulk allocation of mbufs and re-use them.
> > 
> > Also, simplify the overall rx-handler, and update its logic to fix rx perf.
> 
> For such change, it would be nice to provide some benchmark numbers.
> Thanks

On my setup having T580-CR card which is 2-port 40G, I see Rx PPS improving for
64 byte size from ~25Mpps to ~37Mpps on single port.  Similarly, roughly 10
Mpps improvement is seen for dual port also.

And for IO size 128 bytes, approx. 3Mpps improvement is seen.
Of course, my setup is not a powerful one as used by our QA team.
Nevertheless, the improvement is visible in my setup also.

Thanks,
Rahul